AB2013 — The speaker repeatedly references AB 2013 as a California example in a broader discussion of transparency policy for training data disclosures. The bill is used to illustrate how such disclosure requirements might work, how an EU-style template could support implementation, and how enforcement of AB 2013 and related requirements should be understood.
AB2013 — Cited as existing law requiring high-level summaries of training datasets starting January 1, 2026.
AB2013 — Described as a mandatory data disclosure scheme already being led by an Assembly Member Erwin.
Rebecca Bauer-Kahan — The speaker thanks Assembly Member Bauer-Kahan, recognizing her experience, passion, and role as one of the chairs.
Rebecca Bauer-Kahan — Armand Goharbin says AB 412 is sponsored by Assembly Member Bauer-Kahan.
Dean Paul Brest — The speaker says they worked with Dean Paul Brest for a summer and are in Paul Brest Hall.
Mark Lemley — Mark Lemley is mentioned as one of the panelists who started Cairn for Law, the company being discussed, and the speaker notes that it spun out a legal transparency company they ran.
Ian Higgins — The speaker says Ian Higgins from their team is present and hopes attendees meet him.
Professor Samuelson — Professor Samuelson is introduced as the first panelist and identified as being from the University of California, Berkeley, with the committee preparing to ask him the first question.
Professor Samuelson — The chair thanks Professor Samuelson and transitions to the next witness.
Professor Samuelson — Referenced as a Stanford professor whose comments supported fair use and discussed preemption of AB 412.
Ed Lee — Ed Lee is introduced as the speaker’s colleague who is present in the room and is credited with operating chatgptiseatingtheworld.com, a website that compiles and tracks AI litigation materials.
AB412 — The speaker says the bill is likely to be struck down as conflicting with federal copyright law, though states may still regulate deepfakes, privacy, and safety issues.
AB412 — Referenced as a well-intended bill that the witness believes would interfere with the policy approach being discussed.
AB412 — AB 412 is described as a modest but important first step that would require basic transparency for generative AI datasets and help return ownership and recognition to human creatives.
AB412 — AB 412 is discussed as a broadly supported bill aimed at transparency, guardrails, and protection for rights holders and workers in the context of AI and unauthorized use of likeness. Speakers describe it as a needed response to current harms, note support from organizations and individuals, and urge that California enact it as written. The discussion also touches on related concerns such as fair use, possible preemption, and the bill’s role in preventing AI monetization without consent.
Mr. Bomasani / Bonasani — The chair introduces the next speaker; the name appears with uncertain spelling.
Rishi Vamsani — Introduces himself as a senior research scholar at Stanford's Institute for Human-Centered Artificial Intelligence and says he studies frontier AI, its economic effects, and evidence-based public policy.
Mr. Bomasani — The speaker is thanked by name, but the transcript likely refers to Rishi Vamsani/Bomasani from the prior section.
Gail Pellerin — The chair or speaker appears to be turning the discussion to Member Pellerin.
Gail Pellerin — Assembly Member Gail Pellerin is first jokingly described as speechless, then noted as no longer speechless and ready to ask a question, and later thanked by the speaker in the ongoing discussion about watermarking and AI transparency.
Mr. Bomassani — The co-chair references his testimony about data sources and acquisition methods.
Jason George — Jason George testifies on behalf of SAG-AFTRA members about the need for transparency and consent in AI training. He argues that companies should not profit from performers’ work, appearance, or voice without permission and compensation, and explains through a career-stages example how AI can create a substitute performer that mimics his looks, speech, and style. He says that without transparency creators cannot prove their work was used to train AI, leaving them without recourse, and contends that unregulated AI training harms employment and intellectual property rights. George calls for a right to know when creative works are used for AI training, saying transparency would help creators negotiate licenses, enforce rights, and bargain effectively. He notes that licensing for training is feasible and already occurring, and adds that transparency is valuable whether training is licensed or treated as fair use. He closes with the example of an AI-generated song associated with Blanco Brown to show how AI can reinforce stereotypes and extract value from artists’ identities, arguing that transparency could help artists seek justice, compensation, or credit and that legislation is needed.
Jason George — A friend of the speaker who wrote a Korean project that was dubbed into English.
Jason George — Described as an actor and organizer who articulated creators' concerns.
Jason George — Named as an example person whose photo was found in a dataset.
Jason George — The speaker affiliates her comments with Jason George, with the witness being referenced again in the same brief segment.
Jason George — Jason George is mentioned as a point of alignment by multiple speakers, with Amy Heinschike and Andrea Stewart both indicating that their comments are aligned with his and related individuals.
Danny Lynn — Introduced as president of the Animation League.
Danny Lynn — Danny Lynn testifies as a storyboard artist in Los Angeles and president of IATSE Local 839, while clarifying he is speaking in a personal capacity rather than officially for the guild. He explains that animation work is typically short-term and depends on maintaining an online portfolio of storyboard samples to find future jobs. Because of that necessity, he says he has had to assume that any artwork posted online may be scraped into generative AI datasets without his consent or compensation. He argues that generative AI is being marketed to shrink costs and increase productivity, but in practice it is being used to reduce hiring, shorten production schedules, and replace creative labor. He also says AI storyboard tools are low quality and that animation work has been stolen to train replacement technology, concluding that the industry cannot wait until future contract negotiations to address the disruptive impact of AI, especially for non-union workers who lack protections.
Danny Lynn — Multiple speakers reference Danny Lynn in quick succession, with one speaker saying they also agree with Danny Lynn and another asking to affiliate her comments with Danny Lynn. Together, the mentions indicate brief support or alignment with Danny Lynn rather than separate substantive discussions.
Danny Lynn — Referenced as someone whose comments Amy Heinschike aligns with.
Mark Gray — Introduced as Copyright Policy Council at OpenAI and former Assistant General Counsel in the U.S. Copyright Office.
Mark Gray — Mark Gray, Copyright Counsel at OpenAI, introduces himself and begins his testimony on artificial intelligence and copyright, including a brief description of his background at the U.S. Copyright Office.
Mr. George — The chair thanks Mr. George for his testimony.
Mr. George — The discussion centered on Mr. George’s comments about collaboration and partnerships in the media/creative space. The exchange moved from asking about his collaborations, to emphasizing that workers and creatives—not just companies or labels—should be represented in negotiations, to Mr. George explaining that the publicly announced partnerships are mainly with news publishers and are negotiated on mutually agreeable terms. The thread concluded with praise for his point that collaboration must include the actual workers involved, keeping them at the table in these deals.
Mr. Gray — The committee returns to questions for Mr. Gray, who is thanked and asked how the industry can operationalize concerns about protecting creatives without stifling AI or other technology. Mr. Gray responds that he agrees with the concern and outlines several points: partnerships between AI companies, rights holders, and creative sectors are already increasing; the technology is still evolving; historical parallels with earlier technologies like photography suggest similar questions have arisen before; the long-term relationship between technology and the creative sector is hard to predict; and federal copyright is not the right lever to address the issue.
Mr. Gray — The committee thanked Mr. Gray for being present and recognized his expertise as a copyright lawyer, acknowledging his contribution to the discussion.
AB1138 — AB 1138 is referenced twice as an example of legislation supporting the creative sectors, with the discussion emphasizing its role in creative-sector policy and related digital replica issues.
Sam Altman — Referenced as having said AI has both catastrophic risks and immense benefits, and that guardrails are needed.
Luz Rivas — Referenced for work on election integrity and the importance of fact in democracy.
Blanco Brown — Blanco Brown is used as an example of how transparency around model training could help a performer identify that a model was based on his work, allowing him to credibly claim authorship and potentially advance his career.
Mr. Rundberg — The chair indicates the discussion will return to Mr. Rundberg from the next panel.
Mr. DeGraf — Invited to come forward for the next panel.
Professor Powell — Invited to come forward for the next panel.
Professor Powell — Introduced herself and described her background in IP and tech law, offering global perspective on California's options.
Professor Powell — Invited to comment before the committee moved to the next panel.
Professor Powell — The speaker continues technical testimony, explaining that machine unlearning is unreliable and introducing the concept of data entanglement in model training.
Professor Powell — The speaker explains that removing a single concept from a trained model is not cleanly possible and can either fail or damage nearby concepts.
Professor Powell — The speaker says colleges are closing departments because young people see mentors and heroes losing jobs, and cites Chicago and Philadelphia as examples.